Can a 70-year-old start a new career with AI?
Where AI replaces human work, older workers leave their jobs sooner. Where it adds to human work, experience gains value. The future of older people at work depends on that choice.
In Quebec, the average retirement age reached 64.7 in 2023. Until 2030, more people will retire than will enter the labour market. A 60-year-old who planned to slow down can now work with tools that no earlier generation had.
A study by Erik Brynjolfsson, Danielle Li and Lindsey Raymond, published by the National Bureau of Economic Research, shows how powerful these tools are. The researchers followed 5,179 customer service agents. With an AI assistant, they resolved 14% more issues per hour on average. Among novices, the gain reached 34%. Among the most experienced agents, the effect was minimal.
This result raises a question for leaders. If AI gives a beginner what experience used to take months to build, where do older workers fit? Employment data from the past three years gives two answers, depending on how a company uses AI.
AI gives beginners the knowledge of the best.
Two months are worth six. In the NBER study, an agent with two months on the job and the assistant performed as well as an agent with more than six months and no assistant. Less skilled agents even began to write like the most skilled ones.
That knowledge comes from experienced employees. The researchers explain that the tool learned the best practices of the top agents. Part of those practices is tacit knowledge, what a person knows how to do without being able to explain it. AI makes that knowledge available to everyone.
One person can be worth a team. In a study published in 2025, Harvard researchers report an experiment with 776 professionals at Procter & Gamble. One person working with AI matched the performance of a two-person team working without it. Research specialists and commercial specialists both produced more complete solutions, whatever their professional background.
Where AI replaces people, older workers leave sooner.
Older workers are leaving exposed jobs. In June 2026, the Center for Retirement Research at Boston College compared workers aged 55 and over before and after the arrival of ChatGPT. In jobs highly exposed to AI, more of them are leaving work, mostly into unemployment. Among programmers, the increase in departures exceeds 25%.
Few older workers use AI. The same brief cites a survey in which 18% of workers aged 50 to 64 use generative AI.
In Quebec, the main risk lies in prejudice. In October 2025, the Institut du Québec showed that workers over 55 hold 24% of the jobs at risk of automation, a share equal to their weight in the labour force. The economist Souleima El-Achkar sees one of the major challenges in employer prejudice:
The same report notes that 44% of these workers would extend their careers if their employer offered them training after 50.
Where AI adds to people, experience gains value.
Young workers are losing ground and older workers are gaining it. Stanford University's Digital Economy Lab tracks payroll data from the firm ADP. In the occupations most exposed to AI, employment of 22- to 25-year-olds fell about 11% between November 2022 and June 2026. Over the same period, employment in every older age group grew.
Everything depends on how AI is used. The researchers distinguish two uses of AI. When it replaces human tasks, employment falls. When it complements human work, employment holds steady or rises, especially for experienced workers. In occupations that rely on tacit knowledge, employment of the most experienced workers even grows faster. The authors note that these are observations and that they cannot yet prove the cause.
Experience becomes leverage. The researchers offer an explanation. AI more easily replaces knowledge learned from books, the kind a beginner has. It complements knowledge drawn from experience, and so it extends the reach of senior workers.
The exchange between the two intelligences runs both ways.
These studies tell one story. AI carries the knowledge of humanity. People add their judgment, their experience and their intention. Together, the two intelligences go further than either one alone.
Older workers bring what AI cannot find in any text: knowledge of a client, the memory of a crisis, a feel for the trade. Young workers bring ease with the tools. Today, 82% of workers aged 50 to 64 do not yet use generative AI. The elder teaches the intern, and the intern teaches the elder.
A former car exporter became an app developer. In Japan, Tomiji Suzuki built his career at a trading house. In retirement, in the early 2010s, he took a programming course. By June 2024, at 89, he had released 11 free iPhone apps, all designed for older people. For the most recent one, he asked ChatGPT about 1,000 questions. He calls it a "great teacher".
His age is an advantage. The idea for one of his apps came to him the day he forgot his dentures just as he was about to board a train. In his view, "no matter how hard they try", younger people do not fully understand the needs of the elderly. His first career also helps him question the AI. In the days of telegrams, a message had to be clear and fit in one short sentence.
In Japan, he is far from alone. In 2025, 36.2% of Japanese people aged 70 to 74 had a job. That is more than one in three. Among people aged 75 and over, the rate reached 12.6%. Both rates are the highest ever recorded, according to the Statistics Bureau of Japan.
Work by older people also benefits the young. In 2009, the gerontologist Hiroko Akiyama, of the University of Tokyo, created jobs for retirees in Kashiwa, near Tokyo, in their own neighbourhood. According to La Presse, one retiree taught business English to students there. A former engineer started a robotics club. According to the researcher, the health of the retirees who were hired improved.
The young person who is starting out and the older person who is starting over receive the same superpowers.
What decisions do leaders face?
These facts leave companies and governments with three decisions.
- Will AI replace people or work with them? The Stanford data shows that this choice shapes employment. A company can design its tools to complement the judgment of its experienced employees.
- Who trains experienced workers to use AI? Fewer than one in five workers aged 50 to 64 use it. Nearly half of Quebec workers over 55 would stay longer with training after 50.
- How should the exchange between generations be organized? Mentoring can run both ways. A young person can show the tool, and an older person can show the trade.
At Life & Logic, we believe that the blend of the two intelligences is the answer. Someone who planned to retire can then start a new career, with the same superpowers as a young person who is just beginning.
- Generative AI at Work, Erik Brynjolfsson, Danielle Li et Lindsey R. Raymond, National Bureau of Economic Research (document de travail 31161), 2023-04, révisé 2023-11 ; texte intégral
- The Cybernetic Teammate, Fabrizio Dell'Acqua et coll., National Bureau of Economic Research (document de travail 33641), 2025-04
- Are the Careers of Older Workers Being Cut Short by AI?, Geoffrey T. Sanzenbacher, Center for Retirement Research at Boston College (note 26-13), 2026-06
- Les enjeux cachés de la deuxième mi-temps, Institut du Québec et CC45+, communiqué, 2025-10-08
- Canaries in the Coal Mine, version d'août 2026, Erik Brynjolfsson, Bharat Chandar et Ruyu Chen, Stanford Digital Economy Lab, 2026-08 ; résumé
- ChatGPT a mentor for Japan's 89-year-old app developer, Kyoko Hasegawa, Agence France-Presse (repris par The Star), 2024-06-12
- Statistical Topics No. 149, les aînés du Japon vus par la statistique, Bureau de la statistique du Japon (ministère des Affaires intérieures et des Communications), 2026-09-20 ; chapitre sur l'emploi
- Travailler pour rester « dans la vie », Nathalie Collard, La Presse, dossier « La deuxième vie des Japonais », 2026-09-27